🎓 Lesson 6
D5
Queuing Theory for Loading Zones & Dump Points
Queuing theory helps engineers figure out how many trucks should wait at a loading or dumping spot so that no one waits too long—and no equipment sits idle.
🎯 Learning Objectives
- ✓ Calculate steady-state queue metrics (Lq, Wq, ρ) for single-server and multi-server loading zones
- ✓ Design optimal shovel–truck fleet configurations using M/M/c queuing models and utilization thresholds
- ✓ Analyze system sensitivity to arrival rate changes (e.g., due to traffic delays or shovel downtime)
- ✓ Apply Little’s Law to validate field-measured cycle time and queue data
- ✓ Explain trade-offs between capital cost (adding a second shovel) and operational cost (truck idling) using queuing cost models
📖 Why This Matters
In open-pit mines, 30–40% of total haul truck operating cost comes from idle time—much of it spent waiting at shovels or dumps. A poorly sized loading zone can cause cascading delays: trucks queue → shovel cycles stall → fuel consumption spikes → production drops. Queuing theory transforms this intuitive frustration into quantifiable, designable parameters—turning 'how many trucks should we assign?' from guesswork into engineering.
📘 Core Principles
Queuing systems are classified by notation (e.g., M/M/1), where the first letter denotes arrival process (M = Markovian/Poisson), the second service process (M = exponential service time), and the number indicates servers (1 = single shovel). Real mining systems often approximate M/M/c (c identical shovels/dump points) because truck arrivals are nearly Poisson over 15–30 min intervals, and shovel cycle times exhibit sufficient randomness to justify exponential assumptions. Key assumptions include steady-state operation, infinite queue capacity (valid when queues rarely exceed 5 trucks), and first-come-first-served discipline. As utilization (ρ = λ/μc) approaches 1.0, queue length and wait time explode non-linearly—a critical insight for avoiding 'traffic jam' conditions.
📐 Key Calculation
The M/M/c model gives exact formulas for average queue length (Lq) and average waiting time in queue (Wq) when system utilization ρ < 1. These enable direct comparison of fleet sizing alternatives before field trials.
Average Queue Length (M/M/c)
Lq = [ρ^(c+1) / (c! (1 − ρ)^2)] × P₀Expected number of trucks waiting in line for service at a c-server facility (e.g., c shovels).
Variables:
| Symbol | Name | Unit | Description |
|---|---|---|---|
| ρ | System utilization | dimensionless | ρ = λ / (c·μ); must be < 1 for stability |
| c | Number of identical servers | count | e.g., number of shovels or active dump lanes |
| P₀ | Probability of zero trucks in system | dimensionless | Computed from summation formula involving ρ and c |
Typical Ranges:
Stable single-shovel loading zone: 0.65 – 0.82
Optimized multi-shovel dump point: 0.70 – 0.85
💡 Worked Example
Problem: A copper mine uses one hydraulic shovel (c = 1) serving an 8-truck fleet. Measured truck arrival rate (λ) = 4.2 trucks/hour; shovel service rate (μ) = 5.0 trucks/hour. Calculate average number of trucks waiting (Lq) and average wait time (Wq).
1.
Step 1: Compute utilization ρ = λ / (c·μ) = 4.2 / (1 × 5.0) = 0.84
2.
Step 2: Since c = 1, use M/M/1 formula: Lq = ρ² / (1 − ρ) = (0.84)² / (1 − 0.84) = 0.7056 / 0.16 = 4.41 trucks
3.
Step 3: Apply Little’s Law: Wq = Lq / λ = 4.41 / 4.2 ≈ 1.05 hours (63 minutes)
Answer:
The result is Lq = 4.41 trucks and Wq = 63 minutes — indicating severe congestion. Industry best practice targets Wq < 5 min; this system requires either reducing λ (fewer trucks) or increasing μ (faster shovel, e.g., via operator training or bucket upgrade).
🏗️ Real-World Application
At BHP’s Escondida mine (Chile), queuing analysis revealed that adding a second shovel at the primary crusher dump point reduced average truck wait time from 11.2 min to 1.8 min—despite only a 12% increase in capital cost—yielding $2.3M/year in diesel and maintenance savings. The model used 3-month GPS cycle-time logs (λ = 6.8 trucks/hour, μ = 7.5 trucks/hour per shovel) and validated predictions within ±8% of observed queue lengths using M/M/2 with Erlang-C corrections for finite truck population.
📋 Case Connection
📋 Canadian Gold Mine: Steep Ramp Optimization in Narrow Vein Underground
Excessive truck cycle times and premature tire/brake wear due to suboptimal ramp gradient (15%) combined with tight hori...
📋 South African Platinum Mine: Waste Dump Reclaim Optimization
Inefficient haulage routing and underutilized fleet capacity during waste dump reclamation, resulting in excessive diese...